Errata for XQuery 1.0 and XPath 2.0 Formal Semantics

23 November 2007

Latest version:: http://www.w3.org/XML/2007/qt-errata/xquery-semantics-errata.html
Editor:: Michael Dyck, jmdyck@ibiblio.org

Abstract

This document addresses errors in the XQuery 1.0 and XPath 2.0 Formal Semantics Recommendation published on 23 January 2007. It records all errors that, at the time of this document's publication, have solutions that have been approved by the XSL Working Group and/or the XML Query Working Group. For updates see the latest version of that document.

The errata are numbered, classified as Substantive, Editorial, or Markup, and are listed in reverse chronological order of their date of origin. Each entry contains the following information:

A description of the error.
A reference to the Bugzilla entry recording the original problem report, subsequent discussion, and resolution by the Working Group.
Key dates in the process of resolving the error.
Where appropriate, one or more textual changes to be applied to the published Recommendation.

Colored boxes and shading are used to help distinguish new text from old, however these visual clues are not essential to an understanding of the change. The styling of old and new text is an approximation to its appearance in the published Recommendation, but is not normative. Hyperlinks are shown underlined in the erratum text, but the links are not live.

A number of indexes appear at the end of the document.

Substantive corrections are proposed by the XSL Working Group and/or the XQuery Working Group (both part of the XML Activity), which have consensus that they are appropriate; they are not to be considered normative until approved by a Call for Review of Proposed Corrections or a Call for Review of an Edited Recommendation.

Please report errors in this document using W3C's public Bugzilla system (instructions can be found at http://www.w3.org/XML/2005/04/qt-bugzilla). If access to that system is not feasible, you may send your comments to the W3C XSLT/XPath/XQuery public comments mailing list, public-qt-comments@w3.org. It will be very helpful if you include the string [FSerrata] in the subject line of your report, whether made in Bugzilla or in email. Each Bugzilla entry and email message should contain only one error report. Archives of the comments and responses are available at http://lists.w3.org/Archives/Public/public-qt-comments/.

Status of this Document

This is a public draft. None of the errata reported in this document have been approved by a Call for Review of Proposed Corrections or a Call for Review of an Edited Recommendation. As a consequence, they must not be considered to be normative.

Errata

FS.E020 Correct errors in examples in section 2.4.4.

FS.E019 Correct a normalization error in section 4.3.2 Filter Expressions.

FS.E018 Correct various errors in section 8.2.3.1.1.

FS.E017 Update the "Processing Model" diagram.

FS.E016 For internal function fs:idiv, correct the operand types and supporting op: function given in the Binary Operators table.

FS.E015 Fix some small errors in section 8.4.

FS.E014 Fix small errors in section 7.2.3.

FS.E013 Fix small errors in sections 7.2.1 and 7.2.2.

FS.E012 Static Type Analysis must allow for empty text nodes introduced during normalization of direct element constructors.

FS.E011 Correct/simplify/complete the Normalization rules defining []ElementContent and []AttributeContent, and simplify the corresponding Dynamic Evaluation rules.

FS.E010 Clean up due to the removal of op:anyURI-equal from the F+O spec.

FS.E009 Fix some errors in the productions for FunctionSig and TypeList.

FS.E008 Fix a bug in the normalization of function calls.

FS.E007 Fix miscellaneous small errors in section 5.

FS.E006 Fix problems involving the domains of statEnv.funcType and dynEnv.funcDefn: Although section 3.1.1 declares the domain of statEnv.funcType to be (expanded-QName, arity), some inference rules neglect the arity.

FS.E005 Put 'Expr' in parentheses when it occurs as a function argument.

FS.E004 Fix typo in function signature.

FS.E003 Complete the changes entailed by adding a local-namespaces component to the Core CompElemConstructor.

FS.E002 Both Core and Formal grammars had symbols 'NamespaceBinding' and 'NamespaceBindings'.

FS.E001 Make normalization of DirAttributeValue more explicit.

Indexes

Index by affected section

Index by Bugzilla entry

Index by section

Index by production

FS.E020 - editorial

See Bug 4766

Description

Correct errors in examples in section 2.4.4.

History

26 Jun 2007: Proposed

26 Jun 2007: Accepted

4 Oct 2007: Issued

Changes

In 2.4.4 Top level definitions / Example 1, replace:

                  define type UKAddress extends ipo:Address {
                    attribute exportCode of type ipo:UKPostcode
                    element postcode of type positiveInteger
                  };

with:

                  define type UKAddress extends ipo:Address {
                    attribute exportCode of type positiveInteger,
                    element postcode of type ipo:UKPostcode
                  };

In 2.4.4 Top level definitions / Example 6, replace:

A type declaration with one element name of type xs:string follows by one or more elements street of type xs:string.

                  define type Address {
                    element name of type xs:string,
                    element street of type xs:string*
                  }

with:

A type declaration with one element name of type xs:string followed by one or more elements street of type xs:string.

                  define type Address {
                    element name of type xs:string,
                    element street of type xs:string+
                  }

In 2.4.4 Top level definitions / Example 7, replace:

                  define type USAddress extends Address {
                    element zip name of type xs:integer
                  }

with:

                  define type USAddress extends Address {
                    element zip of type xs:integer
                  }

FS.E019 - substantive

See Bug 4578

Description

Correct a normalization error in section 4.3.2 Filter Expressions.

History

24 May 2007: Proposed

30 May 2007: Accepted

4 Oct 2007: Issued

Change

In 4.3.2 Filter Expressions / Normalization / rule 2, replace:

let $fs:sequence := [PrimaryExpr PredicateList]_Expr return

fn:subsequence($fs:sequence, $fs:last,1)

with:

let $fs:sequence := [PrimaryExpr PredicateList]_Expr return

let $fs:last := fn:count($fs:sequence) return

fn:subsequence($fs:sequence, $fs:last,1)

FS.E018 - substantive

See Bug 4242

See Bug 4261

See Bug 4512

See Bug 4581

See Bug 5129

See Bug 3269

Description

Correct various errors in section 8.2.3.1.1.

History

28 Apr 2007: Proposed

3 May 2007: Accepted

3 Oct 2007: Issued

Changes

The following 6 changes resolve bugs 4242, 4261(a), and 4581 (by adding a "?" at the end of the result types), and some of 3269 (by fixing the input types in four of the rules).

In rule 2 / conclusion, replace:

statEnv |- test QName₂ with element of element * OptTypeSpecifier : element QName₂ OptTypeSpecifier

with:

statEnv |- test QName₂ with element of element * OptTypeSpecifier : element QName₂ OptTypeSpecifier ?

In rule 5 / conclusion, replace:

statEnv |- test *:LocalPart₂ with element of element * OptTypeSpecifier : element *:LocalPart₂ OptTypeSpecifier

with:

statEnv |- test *:LocalPart₂ with element of element * OptTypeSpecifier : element * OptTypeSpecifier ?

In rule 7 / conclusion, replace:

statEnv |- test Prefix₂:* with element of element * OptTypeSpecifier : element Prefix₂:* OptTypeSpecifier

with:

statEnv |- test Prefix₂:* with element of element * OptTypeSpecifier : element * OptTypeSpecifier ?

In rule 10 / conclusion, replace:

statEnv |- test QName₂ with attribute of attribute * OptTypeReference : attribute QName₂ OptTypeReference

with:

statEnv |- test QName₂ with attribute of attribute * OptTypeReference : attribute QName₂ OptTypeReference ?

In rule 13 / conclusion, replace:

statEnv |- test *:LocalPart₂ with attribute of attribute * OptTypeReference : attribute *:LocalPart₂ OptTypeReference

with:

statEnv |- test *:LocalPart₂ with attribute of attribute * OptTypeReference : attribute * OptTypeReference ?

In rule 15 / conclusion, replace:

statEnv |- test Prefix₂:* with attribute of attribute * OptTypeReference : attribute Prefix₂:* OptTypeReference

with:

statEnv |- test Prefix₂:* with attribute of attribute * OptTypeReference : attribute * OptTypeReference ?

The following 3 changes resolve bugs 4261(b), 4261(c), and 5129, respectively.

In rule 4, delete premises 2 and 3:

statEnv.namespace(Prefix₁) = (NamespaceKind,AnyURI)

fn:namespace-uri-from-QName(expanded-QName₁) = AnyURI

In rule 6 / premise 2, replace:

statEnv.namespace(Prefix₁) = (NamespaceKind,AnyURI)

with:

statEnv.namespace(Prefix₂) = (NamespaceKind,AnyURI)

In rule 14 / premise 2, replace:

statEnv |- QName₁ of elem/type expands to expanded-QName₁

with:

statEnv |- QName₁ of attr expands to expanded-QName₁

The following 7 changes resolve bug 4261(d).

Between rules 1 and 2, insert this rule:

statEnv |- QName₁ of elem/type expands to expanded-QName₁

statEnv |- QName₂ of elem/type expands to expanded-QName₂

not( expanded-QName₁ = expanded-QName₂ )

statEnv |- test QName₂ with element of element QName₁ OptTypeSpecifier₁ : empty

Between rules 4 and 5, insert this rule:

statEnv |- QName₁ of elem/type expands to expanded-QName₁

fn:local-name-from-QName(expanded-QName₁) = LocalPart₁

not( LocalPart₁ = LocalPart₂ )

statEnv |- test *:LocalPart₂ with element of element QName₁ OptTypeSpecifier₁ : empty

Between rules 6 and 7, insert this rule:

statEnv |- QName₁ of elem/type expands to expanded-QName₁

fn:namespace-uri-from-QName(expanded-QName₁) = AnyURI₁

statEnv.namespace(Prefix₂) = (NamespaceKind,AnyURI₂)

not( AnyURI₁ = AnyURI₂ )

statEnv |- test Prefix₂:* with element of element QName₁ OptTypeSpecifier₁ : empty

Between rules 9 and 10, insert this rule:

statEnv |- QName₁ of attr expands to expanded-QName₁

statEnv |- QName₂ of attr expands to expanded-QName₂

not( expanded-QName₁ = expanded-QName₂ )

statEnv |- test QName₂ with attribute of attribute QName₁ OptTypeReference : empty

Between rules 12 and 13, insert this rule:

statEnv |- QName₁ of attr expands to expanded-QName₁

fn:local-name-from-QName(expanded-QName₁) = LocalPart₁

not( LocalPart₁ = LocalPart₂ )

statEnv |- test *:LocalPart₂ with attribute of attribute QName₁ OptTypeReference : empty

Between rules 14 and 15, insert this rule:

statEnv |- QName₁ of attr expands to expanded-QName₁

fn:namespace-uri-from-QName(expanded-QName₁) = AnyURI₁

statEnv.namespace(Prefix₂) = (NamespaceKind,AnyURI₂)

not( AnyURI₁ = AnyURI₂ )

statEnv |- test Prefix₂:* with attribute of attribute QName₁ OptTypeReference : empty

Delete rule 17:

Lastly, if none of the above rules holds, then the type of the input expression is empty.

ElementNameOrWildcard₁ = NameTest

TypeSpecifier₁ = of type xs:anyType

statEnv |- not(element ElementNameOrWildcard₁ TypeSpecifier₁ <: element ElementNameOrWildcard₂ TypeSpecifier₂)

statEnv |- not(element ElementNameOrWildcard₂ TypeSpecifier₂ <: element ElementNameOrWildcard₁ TypeSpecifier₁)

statEnv |- TypeSpecifier₁ expands to Type₁

statEnv |- TypeSpecifier₂ expands to Type₂

statEnv |- not(Type₁ <: Type₂)

statEnv |- not(Type₂ <: Type₁)

statEnv |- test NameTest with element of element ElementNameOrWildcard₂ TypeSpecifier₂ : empty

The following 2 changes resolve bug 4512(a).

In rule 8 / conclusion, replace:

statEnv |- test * with element of element QName OptTypeSpecifier : element QName OptTypeSpecifier

with:

statEnv |- test * with element of ElementType : ElementType

In rule 16 / conclusion, replace:

statEnv |- test * with attribute of attribute QName OptTypeReference : attribute QName OptTypeReference

with:

statEnv |- test * with attribute of AttributeType : AttributeType

The following 2 changes resolve bug 4512(b).

Between rules 8 and 9, insert this rule:

not( Type is an element type )

statEnv |- test NameTest with element of Type : empty

Between rule 16 and 17, insert this rule:

not( Type is an attribute type )

statEnv |- test NameTest with attribute of Type : empty

The next 2 changes resolve some more of bug 3269.

Delete rule 3:

statEnv |- test QName₂ with element of element OptTypeSpecifier : element QName₂ OptTypeSpecifier

Delete rule 11:

statEnv |- test QName₂ with attribute of attribute OptTypeReference : attribute QName₂ OptTypeReference

FS.E017 - editorial

See Bug 4371

Description

Update the "Processing Model" diagram.

History

6 Mar 2007: Proposed

6 Mar 2007: Accepted

2 Oct 2007: Issued

Change

In 3.2.1 Processing model, in the diagram, add a box labelled "Modules", and an arrow labelled "(SQ3)" pointing from the "Modules" box to the "Static context" box.

FS.E016 - substantive

See Bug 3847

Description

For internal function fs:idiv, correct the operand types and supporting op: function given in the Binary Operators table.

History

18 Oct 2006: Proposed

14 Feb 2007: Accepted

2 Oct 2007: Issued

Change

In C.2 Mapping of Overloaded Internal Functions / Binary Operators table, replace:

fs:idiv(A, B) xs:integer xs:integer op:integer-div(A, B) xs:integer

with:

fs:`idiv`(A, B)	`xs:integer`	`xs:integer`	op:numeric-integer-divide(A, B)	`xs:integer`
fs:`idiv`(A, B)	`xs:decimal`	`xs:decimal`	op:numeric-integer-divide(A, B)	`xs:integer`
fs:`idiv`(A, B)	`xs:float`	`xs:float`	op:numeric-integer-divide(A, B)	`xs:integer`
fs:`idiv`(A, B)	`xs:double`	`xs:double`	op:numeric-integer-divide(A, B)	`xs:integer`

FS.E015 - editorial

See Bug 1783

Description

Fix some small errors in section 8.4.

History

6 Oct 2006: Proposed

16 Jan 2007: Accepted

2 Oct 2007: Issued

Change

In 8.4 Judgments for FLWOR and other expressions on sequences, replace:

[Definition: A Prime types is a choice of item types].

with:

[Definition: A prime type is a choice of item types.]

FS.E014 - editorial

See Bug 1756

Description

Fix small errors in section 7.2.3.

History

8 Sep 2006: Proposed

16 Jan 2007: Accepted

9 Aug 2007: Proposed

9 Aug 2007: Accepted

2 Oct 2007: Issued

Changes

In 7.2.3 The fn:abs, fn:ceiling, fn:floor, fn:round, and fn:round-half-to-even functions / Static Type Analysis / para 1, replace:

Note that the fact that the type should be the least is cannot expressed by the inference rule notation used here.

with:

Note that the fact that the type should be the least is not expressed by the inference rule notation used here.

and delete:

Instead of writing a separate judgment for each function, we write one rule for the functions whose name is either (FN-URI,"abs"), (FN-URI,"ceiling"), (FN-URI,"floor"), (FN-URI,"round"), or (FN-URI,"round-half-to-even") functions.
In 7.2.3 The fn:abs, fn:ceiling, fn:floor, fn:round, and fn:round-half-to-even functions / Static Type Analysis / rule 1 / premise 5, replace:

{ (FN-URI,"ceiling"), (FN-URI,"floor"), (FN-URI,"round"), (FN-URI,"round-half-to-even") }

with:

{ (FN-URI,"abs"), (FN-URI,"ceiling"), (FN-URI,"floor"), (FN-URI,"round"), (FN-URI,"round-half-to-even") }

FS.E013 - editorial

See Bug 1754

Description

Fix small errors in sections 7.2.1 and 7.2.2.

History

5 Oct 2006: Proposed

16 Jan 2007: Accepted

2 Oct 2007: Issued

Changes

In 7.2.1 The fn:last context function / Dynamic Evaluation / rule 1 / conclusion, replace:

dynEnv |- function (FN-URI,"last")() with types on values yields Value

with:

dynEnv |- function (FN-URI,"last") with types on values yields Value

In 7.2.2 The fn:position context function / Dynamic Evaluation / rule 1 / conclusion, replace:

dynEnv |- function (FN-URI,"position")() with types on values yields Value

with:

dynEnv |- function (FN-URI,"position") with types on values yields Value

FS.E012 - editorial

See Bug 3771

Description

Static Type Analysis must allow for empty text nodes introduced during normalization of direct element constructors.

History

10 Oct 2006: Proposed

19 Oct 2006: Accepted

2 Oct 2007: Issued

Changes

In 4.7.3.1 Computed Element Constructors / Static Type Analysis, replace:

The content expression must return a sequence of nodes with all the attribute nodes before any element, processing-instructions or comment nodes. Note that the type allows text nodes interleaved with attribute nodes in the beginning. Although this results in a looser type checking, this accounts for the possible presence of empty text nodes resulting from normalization of direct element constructors (See [4.7.1 Direct Element Constructors]).

with:

The content expression must return a sequence of nodes with attribute nodes at the beginning.

and in each of the four rules, in the last premise, replace:

(attribute*, (element | comment | processing-instruction)*) & text*

with:

attribute *, (element | text | comment | processing-instruction) *
In 7.1.5 The fs:item-sequence-to-node-sequence function / Static Type Analysis, insert:

If the input sequence contains any attribute nodes, they must precede any other items, with the exception that text nodes are allowed throughout. This exception results in looser type checking, but it is necessitated by the possible presence of empty text nodes introduced by normalization of direct element constructors (see [4.7.1 Direct Element Constructors]).

and in the subsequent rule, replace:

attribute*, (element|document|text|processing-instruction|comment|xs:string|xs:float| ...|xs:NOTATION)*

with:

( attribute*, (element|document|processing-instruction|comment|xs:string|xs:float|...|xs:NOTATION)* ) & text*

FS.E011 - editorial

See Bug 3758

See Bug 3760

Description

Correct/simplify/complete the Normalization rules defining []_{ElementContent} and []_{AttributeContent}, and simplify the corresponding Dynamic Evaluation rules.

History

10 Oct 2006: Proposed

19 Oct 2006: Accepted

1 Oct 2007: Issued

Changes

In 4.7.1 Direct Element Constructors / Normalization, delete:

We distinguish between direct element constructors that contain only one element-content unit and those that contain more than one element-content unit.

and delete rule 3:

[ElementContentUnit₁]_{ElementContent}

fs:item-sequence-to-node-sequence(( [ ElementContentUnit₁]_{ElementContentUnit}))

and after the former rule 4 / now rule 3, insert:

(Note that this rule should be understood to cover the degenerate cases of n=0 and n=1, where the element constructor's content consists of zero or one element-content units.)

In 4.7.1.1 Attributes / Normalization, replace:

As with literal XML elements, we need to distinguish between direct attribute constructors that contain one attribute-content unit and those that contain multiple attribute-content units, because the rule for converting sequences of atomic values into strings is applied to sequences within distinct enclosed expressions. If the direct attribute constructor contains exactly one attribute-content unit, we simply normalize that unit by applying the normalization rule for attribute content units:

[ AttributeContentUnit₁ ]_{AttributeContent}

[AttributeContentUnit₁]_{AttributeContentUnit}

If the direct attribute constructor contains more than one attribute-content unit, we normalize each unit individually and ...

with:

To apply []_{AttributeContent} to zero or more attribute-content units, we normalize each unit individually and ...

and after the former rule 5 / now rule 4, insert:

(Note that this rule should be understood to cover the degenerate cases of n=0 and n=1, where the attribute constructor's content consists of zero or one attribute-content units.)

In 4.7.3.2 Computed Attribute Constructors / Dynamic Evaluation, replace:

Second, the function fs:item-sequence-to-untypedAtomic is applied to the content expression and this function call is evaluated in the dynamic environment. Recall from [4.7.3.2 Computed Attribute Constructors] that during normalization, we do not convert the content of direct attribute constructors that contain one attribute-content unit. This guarantees that useful type information is preserved for static analysis. Since the conversion function fs:item-sequence-to-untypedAtomic was not applied to all attribute constructors during normalization, we have to apply it at evaluation time. (As before, it is possible to elide the application of fs:item-sequence-to-untypedAtomic injected during normalization and the application injected during evaluation.)

with:

Second, the content expression is evaluated in the dynamic environment.

and in rule 1 / premise 1, replace:

dynEnv |- fs:item-sequence-to-untypedAtomic(( Expr)) => AtomicValue

with:

dynEnv |- Expr => AtomicValue

and in rule 2 / premise 4, replace:

dynEnv |- fs:item-sequence-to-untypedAtomic(( Expr₂)) => AtomicValue₂

with:

dynEnv |- Expr₂ => AtomicValue₂

FS.E010 - editorial

See Bug 3670

Description

Clean up due to the removal of op:anyURI-equal from the F+O spec.

History

26 Sep 2006: Proposed

26 Sep 2006: Accepted

30 Sep 2007: Issued

Changes

In C.2 Mapping of Overloaded Internal Functions / Notation 2, replace:

fs:`eq`(A, B)	`xs:anyURI`	`xs:anyURI`	op:anyURI-equal(A, B)	`xs:boolean`
...
fs:`ne`(A, B)	`xs:anyURI`	`xs:anyURI`	`fn:not`(op:anyURI-equal(A, B))	`xs:boolean`

with:

fs:`eq`(A, B)	`xs:anyURI`	`xs:anyURI`	op:numeric-equal(fn:compare(A, B), 0)	`xs:boolean`
...
fs:`ne`(A, B)	`xs:anyURI`	`xs:anyURI`	`fn:not`(op:numeric-equal(fn:compare(A, B), 0))	`xs:boolean`

In 8.5.1 Type Promotion / Semantics, insert:

xs:anyURI can be promoted to xs:string:

statEnv |- xs:anyURI can be promoted to xs:string

FS.E009 - editorial

See Bug 3142

Description

Fix some errors in the productions for FunctionSig and TypeList.

History

21 Apr 2006: Proposed

24 Apr 2006: Accepted

29 Sep 2007: Issued

Change

In 3.1.1 Static Context / Notation and A.2 Formal BNF / Non-Terminals, replace:

[85 (Formal)]	`FunctionSig`	::=	`"declare" "function" expanded-QName "(" TypeList? ")" "as" SequenceType`
[86 (Formal)]	`TypeList`	::=	`SequenceType ("," Type)*`

with:

[85 (Formal)]	`FunctionSig`	::=	`"declare" "function" expanded-QName "(" TypeList? ")" "as" Type`
[86 (Formal)]	`TypeList`	::=	`Type ("," Type)*`

FS.E008 - editorial

See Bug 1680

Description

Fix a bug in the normalization of function calls.

History

17 Jul 2005: Proposed

24 Apr 2006: Accepted

29 Sep 2007: Issued

Changes

In 4.1.5 Function Calls / Normalization / rule 1 / premise 2, replace:

statEnv.typeDefn(expanded-QName) = define type QName₂ AtomicTypeDerivation

with:

statEnv |- not(expanded-QName denotes a constructor function)

In 4.12.5 Constructor Functions, add the subsection:

Notation

Calls to constructor functions are normalized differently from other function calls, so we introduce an auxiliary judgment to detect whether the function being called is a constructor function.

statEnv |- expanded-QNamedenotes a constructor function

This judgment holds when the expanded function name maps to an atomic type in the in-scope schema types.

statEnv.typeDefn(expanded-QName) = define type QName AtomicTypeDerivation

statEnv |- expanded-QName denotes a

Errata for XQuery 1.0 and XPath 2.0 Formal Semantics

23 November 2007

Abstract

Status of this Document

Table of Contents

Description

History

Changes

Description

History

Change

Description

History

Changes

Description

History

Change

Description

History

Change

Description

History

Change

Description

History

Changes

Description

History

Changes

Description

History

Changes

Description

History

Changes

Description

History

Changes

Description

History

Change

Description

History

Changes